Deformation imaging allows the direct interrogation of myocardial performance, expanding the ability to evaluate ventricular function in children with congenital heart disease (CHD), including assessment of the subpulmonary and systemic right ventricle, single ventricles, and morphologically abnormal left ventricles. Deformation imaging has provided new insights into the pathophysiology of CHD and diagnostic and prognostic information not readily available using conventional echocardiography. However, at present, deformation imaging is not widely used in clinical pediatric echocardiography. In this review, the authors address the various techniques, their evolving use in CHD, where they have provided important insights in CHD, pitfalls and current limitations to their use, and suggestions for their integration into the routine clinical armamentarium.
Attention ASE Members:
ASE has gone green! Visit www.aseuniversity.org to earn free CME through an online activity related to this article. Certificates are available for immediate access upon successful completion of the activity. Non-members will need to join ASE to access this great member benefit!
A Brief Introduction to the Assessment of Myocardial Mechanics
Pulsed Doppler tissue imaging (DTI) was first introduced to quantify regional myocardial velocities. However, velocities do not truly represent regional myocardial function, because of cardiac translational motion and segmental interactions. Nevertheless, tissue velocities are central to assessment of diastolic function in adults. After the introduction of color DTI, velocities at different points could be simultaneously measured and velocity gradients calculated, corresponding to regional myocardial strain rate. Temporal integration of strain rate curves yielded myocardial strain. In the early 2000s, many investigators used color DTI to study myocardial function in acquired heart disease. However, angle dependency, extensive offline analysis, poor reproducibility, and the ability to analyze restricted segments for circumferential and radial strain have limited its clinical application. After 2004, deformation imaging became more accessible with the introduction of two-dimensional speckle-tracking echocardiography (STE). STE tracks myocardial reflectors from the two-dimensional image independent of angle. In 2009, three-dimensional strain imaging became clinically available, allowing simultaneous analysis of myocardial deformation in all directions and in all segments. However, three-dimensional strain is limited by poor spatial and temporal resolution, which is especially important in children with fast heart rates.
A Brief Introduction to the Assessment of Myocardial Mechanics
Pulsed Doppler tissue imaging (DTI) was first introduced to quantify regional myocardial velocities. However, velocities do not truly represent regional myocardial function, because of cardiac translational motion and segmental interactions. Nevertheless, tissue velocities are central to assessment of diastolic function in adults. After the introduction of color DTI, velocities at different points could be simultaneously measured and velocity gradients calculated, corresponding to regional myocardial strain rate. Temporal integration of strain rate curves yielded myocardial strain. In the early 2000s, many investigators used color DTI to study myocardial function in acquired heart disease. However, angle dependency, extensive offline analysis, poor reproducibility, and the ability to analyze restricted segments for circumferential and radial strain have limited its clinical application. After 2004, deformation imaging became more accessible with the introduction of two-dimensional speckle-tracking echocardiography (STE). STE tracks myocardial reflectors from the two-dimensional image independent of angle. In 2009, three-dimensional strain imaging became clinically available, allowing simultaneous analysis of myocardial deformation in all directions and in all segments. However, three-dimensional strain is limited by poor spatial and temporal resolution, which is especially important in children with fast heart rates.
Rationale for Deformation Imaging in Congenital Heart Disease
Conventional unidimensional (M-mode) or bidimensional (biplane Simpson’s) measurements of myocardial function require normal ellipsoid left ventricular (LV) geometry and are difficult to apply to the right ventricle and univentricular heart. In practice, these are often assessed subjectively. Although deformation is influenced by ventricular geometry, its calculation does not require geometric assumptions, as does the calculation of ejection fraction by M-mode, for example. Therefore, deformation imaging is useful in CHD, in which ventricular shape is often irregular. Deformation imaging can also quantify regional myocardial function when coronary perfusion is potentially compromised, such as in transplant vasculopathy or Kawasaki disease or after arterial switch or Ross operation ( Figure 1 ). Deformation imaging is also useful to assess synchronous contraction of the ventricles and can potentially identify candidates for resynchronization therapy ( Figure 2 ).
Deformation imaging may be able to detect early myocardial dysfunction before the development of overt or irreversible dysfunction. Patients with Duchenne muscular dystrophy and after anthracycline exposure have abnormal strain when ejection indices are still normal. In adults with aortic stenosis, reduced longitudinal strain predicts myocardial fibrosis and lack of functional recovery after aortic valve replacement. Further studies are needed in children with CHD to confirm whether deformation imaging can be used for the early detection of myocardial dysfunction, an application that could influence management and outcomes.
Which Technique Should Be Used in Congenital Heart Disease?
Myocardial deformation imaging is an evolving technology, which has hampered its clinical application in CHD. There is a lack of industry standardization, with vendor-specific solutions for tissue Doppler and speckle-tracking echocardiographic techniques. Furthermore, software was developed for adults, without available pediatric settings. Many variables influence deformation measurements. Therefore, standardized acquisition and analysis are important for reliable measurements. DTI is angle dependent but is acquired at higher frame rates than STE (>180 vs 60–100 frames/sec). The higher temporal resolution may be advantageous in infants with high heart rates, but to date, no direct comparisons between the two techniques have been made in this population. Frame rates mainly affect strain rate measurements, while strain measurements are less frame rate dependent. Strain measurements are dependent on identification of the end-diastolic and end-systolic frame. Because myocardial deformation is a fast event occurring mainly in the first third of the cardiac cycle, identification of the peak strain rate requires high frame rates. In our comparison between speckle-tracking echocardiographic and DTI-based strain calculations, we found strain rate values to be significantly higher when using DTI-based measurements. We think this is related to the different frame rates between the two techniques. Because STE is based on grayscale imaging, it is more dependent on image quality than DTI. During image acquisition, the optimization of wall definition helps with the quality of speckle-tracking analysis. DTI-derived strain is less dependent on the two-dimensional quality of the images. We used DTI-derived strain in patients with Duchenne muscular dystrophy with good reproducibility for radial strain despite often poor echocardiographic windows. We recently demonstrated that in children (many with CHD), STE reliably measures longitudinal and circumferential strain, but radial strain measurements are highly variable and vendor dependent. Offline strain measurement from stored Digital Imaging and Communications in Medicine images performs reasonably well for longitudinal and circumferential strain but did not improve radial strain reproducibility. Therefore, we currently recommend STE for longitudinal and circumferential strain and DTI for radial strain. This approach limits radial strain analysis to the posterior wall segments in the short-axis or long-axis view. Generally, STE has the advantage of being more angle independent compared to tissue Doppler–derived strain measurements, the only exception being rotation analysis. Reproducibility of deformation analysis is likely more problematic in neonates and preterm infants, mainly because of the higher heart rates and the thinner myocardial walls with less speckles, although some reproducibility data are more reassuring.
Strain rate measures deformation velocity and would theoretically assess ventricular function better. In an animal model, peak systolic strain rate was shown to be less influenced by loading conditions and cardiac structure compared with end-systolic strain measurements. Unfortunately, strain rate measurements are more variable, and improvements in technology are further required. Improving technology to optimize strain rate measurements would be a significant advantage for CHD, in which loading conditions are variable and affect myocardial deformation properties. This is even more important for diastolic measurements. We demonstrated poor reproducibility of early diastolic strain rate measurements, which is likely attributable to the rapid occurrence of early diastolic events relative to the low temporal resolution of STE and even DTI. Therefore, we currently do not recommend early diastolic strain rate measurements for clinical diagnostic purposes.
Another major challenge is defining normal ranges for children ( Table 1 ). For DTI strain measurements, normal data have been published for children and neonates. Serial decrease in LV strain and increase in right ventricular (RV) strain in the first month of life were attributed to afterload changes and myocardial maturation. Normal speckle-tracking echocardiographic strain values in children have recently been published, with wide confidence intervals and discrepant results regarding the effects of age and heart rate. Therefore, further data are required. Normal data on developmental changes in ventricular rotation and twisting as measured by STE have been published. LV torsion increased with age because of increased apical rotation. However, when rotation was normalized for heart size, somatic growth did not affect torsion. Younger hearts appear to twist and untwist faster.
| LV longitudinal | Radial | Circumferential | RV longitudinal | |||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Study | Age | n | Method | Basal | Mid | Apical | Posterior | Anterior | Basal | Mid |
| Weidemann et al. | 4 to 16 y | 33 | CDMI | −26 ± 11 | −26 ± 08 | −25 ± 07 | 58 ± 12 | — | −36 ± 11 | −45 ± 13 |
| Boettler et al. | 0 to 16 y | 124 | CDMI | −21 to 30 (50% CL, 16 to 38) | 21 to 30 (50% CL, 17 to 37) | 19 to 33 (50% CL, 14 to 41) | — | — | — | — |
| Lorch et al. | 0 to 18 y | 284 | VVI | 20.68 ± 8.08 | — | — | — | — | — | — |
| Marcus et al. | 0 to 19 y | 195 | STE | −17 ± 2.6 to 20.4 ± 1.6 | −18.2 ± 2.7 to 22.5 ± 1.4 | 20.4 ± 1.9 to 25.1 ± 1.2 | 63.2 ± 11.6 to 66.8 ± 4.1 | 18.2 ± 1.6 to 22.9 ± 2.0 | ||
| Kutty et al. | 1 to 18 y | 30 | VVI | — | — | — | — | — | 18.9 ± 4.7 | 25.4 ± 4.7 |
| Pena et al. | 1 d | 55 | CDMI | −24.46 ± 3.82 | −24.36 ± 3.53 | −24.40 ± 3.48 | 55.72 ± 12.13 | |||
Effects of Geometry and Loading
Strain reflects regional myocardial deformation and can be considered a regional myocardial “ejection fraction.” In the absence of regional dysfunction, strain measurements correlate well with global ejection fraction. Strain rate is the speed of deformation and reflects the velocity of fiber shortening. Accordingly, it correlates with invasively measured dP/dt and end-systolic elastance. Segment deformation is influenced by intrinsic myocardial force development, external forces (afterload, preload, and segment interaction), and tissue elasticity. Thus, reduced strain may reflect decreased contractility but also increased afterload or reduced preload. Moreover, there is a difference between acute and chronic changes. Acute pressure loading reduces deformation. Chronic pressure loading causes hypertrophy that normalizes afterload and deformation. Acute volume loading is expected to increase strain. Chronic volume loading will increase ventricular volumes because of eccentric remodeling and reduce strain. In aortic and mitral insufficiency, deformation measurements have been corrected for LV volume. Patients below the 95% confidence interval of this relationship are hypothesized to have ventricular dysfunction. The influence of acute and chronic changes in loading conditions on different strain and strain rate measurements requires further study. Understanding the physiologic adaptations in myocardial deformation to changes in loading and chronic remodeling is a prerequisite for understanding pathologic adaptations in CHD. Once the effect of physiologic remodeling on myocardial deformation is better described, pathologic changes can be better distinguished.
The effects of loading might also be different in the right ventricle compared with the left ventricle, with different mechanisms of adaptation and remodeling occurring in the right ventricle.
Deformation Imaging for Atrial Function
Most research on deformation imaging in CHD has focused on ventricular function, although atrial function is important as well. During ventricular contraction, the atria form a reservoir for blood returning from the pulmonary and systemic veins. Atrial filling is enhanced by the apical displacement of the atrioventricular valve annulus during ventricular contraction. Therefore, ventricular deformation also affects atrial filling. After atrioventricular valve opening, the atria form a conduit transferring blood to the ventricles. The atria therefore are affected by ventricular relaxation. Atrial contraction in late diastole contributes substantially to ventricular filling and therefore cardiac output, especially in patients with decreased ventricular function. It is during this phase when deformation imaging is likely most useful to assess atrial booster pump function. However, although atrial function may be important in CHD, it has received relatively little attention. Strain imaging has been used to study atrial contraction after device versus surgical closure of atrial septal defects (ASDs). Both right and left atrial strain have been found to be better preserved after device versus surgical closure of ASDs. This suggests that cardiopulmonary bypass affects atrial as well as ventricular myocardial function, because right atriotomy would not be expected to affect left atrial deformation. Decreased right atrial function by tissue Doppler–derived strain rate has also been demonstrated in children and adults after tetralogy of Fallot (TOF) repair. This population had a wide age range between 0.3 and 51 years. Right atrial strain rate was correlated with decreased atrial emptying, although it was enhanced in patients with low RV ejection fractions. These data need to be interpreted in light of multiple confounders, such as the wide patient age range, the wide range of age at repair, right atrial enlargement, tricuspid and pulmonary regurgitation (PR), RV outflow obstruction, and restrictive RV physiology, which may be present in some patients. The same research group compared atrial strain in Fontan patients with intra-atrial versus extra-atrial baffles. Both ventricular and atrial strain were lower in Fontan patients compared with controls. Moreover, atrial strain was lower in Fontan patients with intra-atrial baffles compared with those with extra-atrial baffles. These findings are interesting, because patients who have undergone Fontan procedures, especially those with older-type Fontan circulations, are susceptible to atrial enlargement and atrial arrhythmias. Whether atrial dysfunction can be followed over time by strain imaging to detect which patients are at risk for developing atrial dysfunction and arrhythmia needs to be determined by further study.
Cardiac Torsion
The opposing basal clockwise and apical counterclockwise rotation of the left ventricle creates a twisting or torsional motion during systole and an untwisting motion during diastole that can be measured using STE. Some studies have suggested that torsion increases over childhood, primarily because of increasing basal rotation. However, when corrected for ventricular length, torsion is stable across heart sizes and age, or even increased in infants. In chronic compensated LV afterload from aortic stenosis or aortic coarctation, torsion is increased, perhaps as a compensatory mechanism. Consequently, after relief of the obstruction, torsion decreases toward normal values. Ventricular interactions may lead to decreased LV torsion and untwisting in conditions of RV afterload, such as TOF. This may be due to changes in LV longitudinal and circumferential strain in these patients or even to decreased RV strain through ventricular interactions, as RV apical strain is correlated with LV apical strain and apical LV rotation. RV adaptation to systemic resistance has been studied in transposition of the great arteries after atrial switch procedures. Although systemic right ventricles adapt a LV contraction pattern with dominant radial strain over longitudinal strain, they do not develop torsion, as a normal left ventricle would. Torsion in single ventricles is also markedly abnormal, with abnormal apical strain and rotation.
Deformation Imaging in Congenital Heart Disease
We now turn our attention to deformation imaging by echocardiography in specific CHD lesions, focusing on ASD, TOF, the systemic right ventricle, the functionally univentricular heart, especially hypoplastic left heart syndrome (HLHS), and coarctation of the aorta.
Deformation Imaging in ASDs
The effect of volume loading on myocardial performance in CHD can be assessed in shunt lesions such as ASDs, which create RV volume loading. In contrast to patients with TOF, RV longitudinal deformation is largely preserved or even elevated in children with ASDs, even though they too have dilated right ventricles. Not only is RV strain generally preserved in children with ASDs, but it remains largely unchanged immediately (within 24 hours) after ASD closure with a device There may be differences, however, between catheter device and surgical ASD closure in terms of their effects on RV function, as demonstrated by deformation imaging. Di Salvo et al. showed that RV (and LV) deformation are better after catheter device closure of the defect compared to surgical ASD closure. Similar results in the same population were obtained with regard to left and right atrial deformation, with atrial strain being similar to controls after device closure but reduced after surgical closure. As previously mentioned, these results mirror previous findings obtained with M-mode echocardiography, in which RV lengthening was better preserved after ASD device closure compared with surgery, possibly reflecting the effects of cardiopulmonary bypass on myocardial performance. However, because the long-term clinical prognosis after successful closure of an ASD with surgery or device in childhood is excellent, the implications of reduced ventricular or atrial strain after ASD closure are uncertain. Whether changes in myocardial performance are associated with long-term risks of ASDs, such as arrhythmia, remains an open question. The lack of change in RV deformation after device closure in children contrasts with results obtained in adults, in whom RV strain decreased after device closure of ASDs. This may reflect the effects of longer standing volume load and their effects on RV deformation, even in relatively young adults and even though preclosure RV strain is generally preserved in the adult population. Strain is related to volume change, and accordingly, strain rate, which better reflects contractility compared with strain, remains relatively unchanged after device closure. Nevertheless, both younger and older patients can experience improved exercise capability after ASD closure, related to RV remodeling and improved RV function. Whether this improvement is related to improved myocardial deformation remains to be proven. However, early data have suggested that regional RV strain, especially RV apical strain, correlates with functional capacity after ASD closure. Higher-than-normal RV apical strain before closure decreased after ASD closure, and postclosure apical strain was correlated with functional capacity.
Although there are some data on tissue velocities in children with isolated ventricular septal defects, predominantly to study the effects of chronic volume loading in CHD and myocardial performance after ventricular septal defect closure, there is a lack of similar data by strain imaging.
TOF
Patients with repaired TOF commonly experience progressive exercise intolerance, often associated with RV dysfunction. Chronic PR leads to progressive RV enlargement and dysfunction, tricuspid regurgitation, ventricular arrhythmias, and sudden death. Deformation imaging may reveal early RV myocardial dysfunction. However, defining relations between regional myocardial dysfunction, PR, RV chamber dysfunction, clinical symptoms, and exercise capacity has proven elusive. It is often difficult to know if reduced segmental strain in an asymptomatic patient is clinically important. RV strain may be decreased even in asymptomatic children with relatively good RV function after TOF repair ( Figure 3 ). The interpretation of the myocardial deformation parameters is complex because of the interaction between different physiologic parameters influencing strain measurements in postoperative patients with TOF. The increased stroke volume caused by PR should increase RV myocardial deformation, while RV dilatation has the opposite effect on myocardial deformation, as a larger ventricle needs to deform less to generate the same output. Concentric hypertrophy with the development of more circumferentially oriented fibers has been described, which could result in a more important circumferential contribution to RV ejection. This is difficult to study using current strain methodology. Finally, pathologic changes within the myocardium related to myocardial hypertrophy and wall fibrosis result in decreased myocardial deformation. This probably explains why different studies performed in different study populations generated different results. In one study, RV size and basal peak systolic strain rate correlated well with QRS duration (a surrogate for RV size), but RV deformation did not correlate with PR severity. Conversely, patients with transannular patches, who typically have more PR, had worse strain. Variable RV dilatation, inexact grading of PR, and heterogeneous RV deformation may all affect results. More recent work confirmed decreased RV longitudinal deformation after TOF repair, but in this study, RV strain was associated with PR, RV ejection fraction, and exercise capacity. Overall, results suggest that chronic PR is associated with RV enlargement and in some patients RV global dysfunction, reduced myocardial regional deformation, and decreased exercise capacity ( Figure 4 ). Regional myocardial function is important after TOF correction and is related to myocardial damage caused by the surgical technique as well as by the remodeling related to adaptation to chronic volume loading. The RV outflow tract is damaged by the surgical patch, with extension of scar tissue into the anterior RV wall, resulting in significant RV outflow tract hypokinesia and dyskinesia. This is difficult to measure using strain measurements, because the RV outflow tract is difficult to visualize. RV remodeling results in cross-sectional dilatation of the RV apex and RV inlet, leading to basal bulging and angulation of the tricuspid annulus. Using cardiac magnetic resonance imaging, this was described to be associated with regional differences in area-strain, with increased area-strain in the RV apex and decreased measurements in the basal and outflow segments. Recent three-dimensional volumetric analysis also indicated apical RV volumes to be the most increased compared with normal controls, with better preservation of the apical ejection fraction. Currently, no good regional echocardiographic data are available, but investigating at the RV inlet, body, and outlet using speckle-tracking strain would be an area of considerable clinical interest. Additionally, it would be of value to be able to study circumferential or radial strain in the different RV segments to assess whether concentric RV hypertrophy contributes to increased radial or circumferential strain.
